CASRAI Dictionary

Tag: CRediT extensions

  • Recognising mentorship and training contributions in the research record

    Ask any senior researcher what they are proudest of, and a striking number will name the people they trained rather than the papers they wrote. Yet mentorship and training are almost entirely invisible in the formal research record. There is no DOI for supervising a doctoral researcher to completion, no citation count for the postdoc you helped launch, no structured field anywhere that records the years of pastoral and intellectual labour that hold a research group together. This is the archetypal hidden labour of research, and a cluster of recent developments is beginning to make it countable. This article surveys them, drawing on the mentorship and career-stages domain.

    Why the gap exists

    The formal record evolved to capture outputs — articles, books, patents, datasets — because outputs are discrete, attributable, and citable. Mentorship is none of those things. It is continuous, diffuse, and its effects show up years later in someone else’s career. The traditional CV gestures at it (“supervised 12 PhD students”) but in a form that is unverifiable, uncomparable, and easy to inflate. The result is a systematic under-recognition of exactly the work that sustains research culture, and a corresponding incentive to neglect it in favour of countable outputs.

    Narrative CVs: making space for the contribution

    The most consequential development is the shift toward narrative CVs. The UK Research and Innovation funder, UKRI, made its Résumé for Research and Innovation (R4RI) format standard across all its funding from January 2024; the Royal Society’s Résumé for Researchers preceded it, and Wellcome and others run comparable formats. These replace the enumerated publication list with a structured narrative organised around contribution types — and, crucially, they explicitly ask researchers to describe their contributions to people and to the research community, not only to knowledge.

    The R4RI structure asks for contributions across several modules, one of which is explicitly about the development of individuals — mentoring, supervision, team-building, and support for others’ careers. For the first time in a mainstream funding format, “I mentored three early-career researchers into independent positions” is not a throwaway line at the bottom of a CV but a first-class, evaluated contribution. The narrative form is what makes this possible: mentorship resists enumeration, but it can be described, and a good narrative description is assessable by a panel in a way that a raw number never was.

    Career-stage vocabulary: the precondition for fair comparison

    Recognising mentorship fairly requires knowing who is being mentored and where they are in their career — which requires a shared career-stage vocabulary. The terms look mundane but their absence causes real unfairness. A doctoral researcher, a postdoctoral researcher, an early-career researcher, a mid-career researcher, an established researcher — these are not interchangeable, and the expectations attached to each differ. Funder definitions of “early-career” vary widely, which means a researcher can be eligible for an ECR scheme in one country and not in another for no principled reason.

    Just as important are the terms for career breaks — parental, caring, illness, military service — and for part-time and fractional working. These exist in the vocabulary for a specific reason: responsible-assessment regimes expect evaluators to make career-stage adjustments, judging a researcher’s track record relative to the time and circumstances they actually had. A researcher who took two years of parental leave and works at 0.6 FTE should not be assessed as though they had a continuous full-time career. None of that is possible without a controlled vocabulary that lets the relevant facts be recorded and read consistently. Career-stage terms are, in this sense, equity infrastructure.

    Recording the mentorship relationship itself

    Beyond the CV, there is the question of recording the mentorship relationship as structured data. The vocabulary distinguishes a primary mentor from a secondary mentor, a thesis supervisor from a postdoc mentor, and records events such as mentee completion — a mentorship reaching a successful conclusion, a degree awarded, a postdoc transitioned to their next position. Where these are captured as structured records, with the people involved identified by ORCID iD, a mentorship history becomes something a researcher can carry with them, claim on a narrative CV, and have verified — rather than an unverifiable assertion.

    CRediT extensions and the limits of the current taxonomy

    How does CRediT handle this? Only partially, and that is a recognised gap. CRediT’s Supervision role covers “oversight and leadership responsibility… including mentorship external to the core team,” which captures mentorship that shapes a specific output. But CRediT applies to outputs, and most mentorship is not attached to a single paper. The doctoral supervision that shaped a researcher over four years is not well described by a Supervision tag on one of their papers.

    This is one of the motivations behind the active work on CRediT extensions and adjacent contribution vocabularies — roles for mentors, technical staff, and other acknowledged contributors whose work the 14-role taxonomy does not capture. The honest position is that mentorship is better served by the narrative CV and by structured relationship records than by stretching the output-level CRediT statement to cover it. CRediT credits contribution to a work; mentorship is contribution to a person, and the field is still building the vocabulary for the latter. Initiatives such as the Hidden REF have done much to make the case that this labour should be visible at all.

    What to do now

    For researchers: use the mentorship and career-development modules of narrative CV formats fully — describe the people you have developed, not just the papers you have produced. For institutions and funders: adopt a consistent career-stage vocabulary, record career breaks and fractional working, and make genuine career-stage adjustments in assessment. For vocabulary work: prioritise the structured representation of the mentorship relationship and the CRediT extensions for acknowledged contributors. The labour that builds the next generation of researchers should be visible in the record that generation inherits.

    Related reading

  • Hyperauthorship and large collaborations: crediting contributions at scale

    Some research papers now carry author lists that run to several pages — thousands of named individuals on a single article, common in particle physics, large-scale genomics, and global clinical consortia. The phenomenon has a name, hyperauthorship, and it puts real pressure on the idea of authorship itself: when the list is that long, what does being “an author” still tell a reader? This article looks at how credit is kept meaningful at that scale, drawing on group-authorship conventions and contributor-role metadata. It connects to the CRediT taxonomy and its fourteen roles, and to the wider group-authorship guidance.

    What hyperauthorship breaks

    The traditional author line works as a signal because it is short enough to read as information: position implies contribution, and a reader can form a view of who did what. Hyperauthorship breaks that signal in two directions at once. The list becomes too long to convey anything about individual contribution — position is meaningless when there are two thousand positions — and at the same time the conventional authorship criteria become hard to apply, because the standard ICMJE requirement that every author be able to take public responsibility for the whole work strains when no single person could possibly have overseen every part of a vast collaboration.

    The result is a genuine tension. Excluding contributors who did essential work would be unjust and would erase the labour that made the result possible. Including everyone as a conventional “author” empties the term of meaning. Large collaborations resolve this not by abandoning credit but by changing how it is structured.

    Consortium and group authorship

    The first mechanism is group (or consortium) authorship. Rather than printing two thousand names on the byline, the article is attributed to a named collaboration — “the SomeStudy Consortium” — with the full list of contributing individuals recorded in a structured way, typically in a collaborator list that is itself indexed so that the individuals remain discoverable and their contribution citable. The named group becomes the author-of-record on the byline, while the individuals are preserved in the metadata.

    This is more than a typographical convenience. It reflects a real fact about the work: the contribution that matters is often the collaboration’s, a collective enterprise with its own governance, membership rules, and authorship policy, sustained across many projects. Treating the consortium as an entity — ideally one that can itself be identified, much as a RAiD identifies a research activity — lets the collective be credited as a collective, while the structured collaborator list ensures individuals are not lost. Indexing services and ORCID support representing membership of such author groups, so an individual can still claim their participation on their own record.

    Contributor roles at scale

    Group authorship answers who was involved; it does not by itself answer who did what. That is where contributorship — recording specific roles rather than a flat “author” label — becomes essential rather than merely nice-to-have. The CRediT taxonomy‘s fourteen roles let a large collaboration record, in structured form, that one group ran the instrument (Investigation, Resources), another built the analysis pipeline (Software, Formal analysis), another curated the shared data (Data curation), and a smaller group drafted the paper (Writing – original draft). The optional degree-of-contribution qualifier — lead, equal, or supporting — adds a further layer, distinguishing those who led a function from the larger number who supported it.

    For a hyperauthored paper, this is the difference between a credit record that is meaningful and one that is not. A flat list of two thousand authors is opaque; the same two thousand contributors mapped to roles — even at the level of teams rather than individuals — tells a reader, a committee, or an indexing system something real about how the work was actually divided. CRediT was designed precisely for this: a small, learnable, structured vocabulary that scales because it describes functions, which remain finite even when contributors do not.

    Practical principles for crediting at scale

    1. Set an authorship and contributorship policy up front. Large collaborations should agree, before results exist, what qualifies someone for the collaborator list, how roles are assigned, and how the consortium is named. Retrofitting credit onto a finished mega-paper is where disputes start.
    2. Use the named-group byline with a structured collaborator list. Attribute the article to the identified collaboration, and record every contributing individual in indexed metadata so they remain discoverable and can claim the work on their ORCID record.
    3. Record roles, not just names. Map contributions to CRediT roles — at team granularity where individual granularity is impractical — so the credit conveys structure, not just headcount.
    4. Make the collaboration itself identifiable. Give the consortium a stable identity (and, where applicable, a project identifier such as RAiD) so its outputs aggregate as a coherent body of work over time.
    5. Keep responsibility explicit. Designate who is accountable for the integrity of the whole and who corresponds — the ICMJE expectation of accountability does not disappear at scale; it has to be assigned deliberately rather than assumed of everyone.

    What stays constant

    It is tempting to treat hyperauthorship as a special case with its own rules, but the underlying principle is the one that governs all credit: contribution should be recorded honestly and in a form that travels. A two-author paper and a two-thousand-author paper face the same question — who contributed what, and how is that preserved so it can be read later — and answer it with the same tools: clear roles, structured metadata, and stable identifiers. Scale changes the mechanics (named groups, indexed collaborator lists, team-level role mapping) but not the goal. Credit at scale is still credit; it just has to be engineered rather than left to the author line.

    Where shared vocabulary fits

    “Author”, “contributor”, “collaborator”, “consortium author”, and “group authorship” are recorded inconsistently across venues, which is exactly how individual contributions to large collaborations get lost. A shared, federated vocabulary that defines these terms precisely — and points back to NISO for the CRediT standard and its degree qualifier, and to ICMJE for the authorship criteria — is what lets a contribution to a mega-collaboration mean the same thing wherever it is read. Supplying that definitional layer is the role the CASRAI dictionary is designed to play; the relevant terms sit in the credit-extensions domain.

    Related reading

  • Recognising peer reviewers: from anonymous service to credited contribution

    Peer review is the labour the scholarly system depends on most and rewards least. Reviewing a manuscript well takes hours of expert attention — reading carefully, checking methods, catching errors, sometimes reshaping a paper substantially — and almost all of it happens anonymously, unpaid, and unrecorded. The reviewer’s name never appears, the report is rarely seen, and the work leaves no trace on any CV. Making that contribution visible and creditable, without necessarily compromising the anonymity that protects candid review, is the problem this article is about. It sits in the credit-extensions domain and connects to the wider question of who gets credit for what, addressed through the CRediT taxonomy.

    Why review is the great uncredited contribution

    The invisibility of review is not an oversight; it is structural. Most review is single- or double-anonymous by design, and for good reason — anonymity lets a reviewer write a frank, critical report without fear of reprisal, particularly when assessing the work of someone more senior. But the same anonymity that protects candour also erases recognition. A researcher who reviews thirty manuscripts a year has nothing to show for it, while the system quietly assumes they will keep doing it. The result is a recognition gap that bears hardest on the early-career researchers who do a great deal of reviewing and have the most to gain from having it counted.

    It is worth being clear about scope. The CRediT taxonomy deliberately covers authorship contributions to a specific paper; it has no role for the reviewers of that paper, because they are not contributors to it in the authorship sense. Recognising review is therefore an adjacent problem to CRediT rather than something CRediT itself solves — which is exactly why a dedicated vocabulary and dedicated infrastructure for reviewer recognition matter.

    The shift from anonymous service to recorded activity

    The key insight behind reviewer recognition is that you can record that a review happened — and credit the reviewer for it — without revealing what the review said or which way it leaned. The unit of recognition is the review activity: a verified record that a named researcher completed a review for a named venue on a given date. The content stays confidential; the contribution becomes visible. This decoupling is what makes it possible to credit review without breaking the anonymity that makes honest review possible in the first place.

    ORCID review records

    ORCID supports peer review as a first-class activity type on a researcher’s record. A journal or platform that integrates with ORCID can deposit a structured review record directly onto the reviewer’s ORCID profile: it states that the person performed a review for a particular organisation, at a particular time, typically without disclosing the manuscript or the verdict. Because the record is asserted by the venue rather than self-claimed, it is verified — it carries the weight of having come from the journal, not merely the reviewer’s say-so. Over time, a reviewer accumulates a trustworthy, machine-readable record of their review activity that travels with their ORCID iD into CVs, funding applications, and institutional systems.

    Web of Science reviewer recognition

    Web of Science reviewer recognition — the service that grew out of the platform formerly known as Publons — provides a complementary route. It lets reviewers build a verified record of their reviewing (and editorial) activity across journals, again typically capturing the fact and venue of a review rather than its confidential content, and presents it as a profile a researcher can point to. Many publishers feed verified review activity into this system automatically, and it interoperates with ORCID so that the same activity can surface on a researcher’s ORCID record. The two together — ORCID as the open, portable identity layer and Web of Science as a recognition platform that aggregates and displays activity — form the practical backbone of reviewer recognition today.

    Open review and stronger forms of credit

    Where a venue practises open peer review, the recognition can go further. If a reviewer chooses to sign their report, or if the report is published alongside the article (with or without the reviewer’s name), the review becomes a citable object in its own right — an output a reviewer can point to directly, not merely an activity record. This is the strongest form of review credit, because it makes not just the fact of the review but its substance part of the public record. It is optional and not appropriate for every context, but where it is used it turns review from invisible service into a visible scholarly contribution. (For the trade-offs of opening review, the distinction between signed and transparent models matters a great deal.)

    Why recognition matters beyond fairness

    Crediting review is not only about being fair to reviewers, though it is that. It also serves the system. A reviewer-recognition record gives editors a verifiable signal of who reviews, how much, and in what fields — useful for finding and acknowledging reliable reviewers. It gives funders and hiring committees, increasingly under responsible-assessment reforms that value contribution over output-counting, a legitimate way to see and reward an activity that crude publication metrics ignore entirely. And by making the labour visible, it makes the implicit bargain of the system explicit: review is work, work deserves recognition, and recognition can be recorded without compromising the confidentiality review depends on. The same principle that animates the credit due to authors applies to reviewers — contribution should be recorded honestly and in a form that travels.

    Where shared vocabulary fits

    “Peer review”, “reviewer recognition”, “review record”, “signed review”, and “review activity” are recorded inconsistently across journals and platforms, which is exactly how review contributions get lost or double-counted. A shared, federated vocabulary that defines these terms precisely — and points back to ORCID’s peer-review schema and the recognised reviewer-recognition platforms — is what lets a review credited in one system be understood in another. Supplying that definitional layer is the role the CASRAI dictionary is designed to play; the relevant terms sit in the credit-extensions domain.

    Related reading

  • Crediting contributions in systematic reviews and meta-analyses

    A systematic review looks, from the outside, like a single coherent document with a tidy list of authors. From the inside it is a small project with a remarkable division of labour: a protocol to register, a search strategy to design and run across multiple databases, thousands of records to screen against eligibility criteria, full texts to retrieve and assess, data to extract twice over, risk-of-bias judgements to make, a synthesis or meta-analysis to compute, and a report to write to an exacting standard. Each of those tasks is a distinct skill, and each is usually done by a different person or pair of people. The conventional author byline flattens all of it. This article looks at how structured reporting through PRISMA and structured contributorship through the CRediT taxonomy together make the real shape of this work visible, and where the vocabulary for it sits in the credit extensions domain of the CASRAI Dictionary.

    Why a review is hard to credit fairly

    The difficulty is that the most laborious and methodologically critical parts of a review are precisely the ones that leave no trace in a traditional byline. Screening twenty thousand abstracts in duplicate is exacting, consequential work — get the eligibility judgements wrong and the whole review is compromised — yet it is invisible in author order. The same is true of designing a reproducible search, performing duplicate data extraction, or making risk-of-bias assessments. Meanwhile, the person who conceived the question and the person who drafted the manuscript are easy to recognise. A fair account of a review has to name the unglamorous, high-stakes tasks as clearly as the visible ones.

    PRISMA: reporting the process transparently

    The first half of the answer is methodological transparency. PRISMA — Preferred Reporting Items for Systematic Reviews and Meta-Analyses — is the reporting guideline that tells readers what a review actually did: how the search was constructed, how records moved from identification through screening to inclusion (the familiar flow diagram), how data were extracted, and how studies were appraised and synthesised. PRISMA does not assign credit, but it makes the work auditable. When a review reports its process to the PRISMA standard, the existence and scale of each task — the searching, the screening, the extraction, the appraisal — becomes explicit rather than implied. That visibility is the precondition for crediting it: you cannot recognise a contribution that the reporting has hidden.

    CRediT: naming who did what

    The second half is contributorship. The Contributor Roles Taxonomy provides a controlled vocabulary of contribution types that maps unusually well onto the anatomy of a review. The full set is set out in our overview of the CRediT roles, but several are worth singling out for evidence synthesis:

    • Conceptualization — formulating the review question and eligibility criteria.
    • Methodology — designing the search strategy and the synthesis approach, often the work of an information specialist.
    • Investigation — running the searches, screening records and retrieving full texts.
    • Data curation — managing the extracted data, de-duplication and the records that underpin the flow diagram.
    • Formal analysis — the meta-analysis itself, including heterogeneity assessment and any sensitivity analyses.
    • Writing – original draft and Writing – review & editing — producing and refining the manuscript.

    Used together, these roles let a review record that the information specialist designed the search, that two named reviewers screened and extracted in duplicate, and that the statistician ran the synthesis — rather than leaving all of it to be guessed from author order. The wider CRediT taxonomy turns the division of labour into a machine-readable statement attached to the output.

    The role information specialists deserve

    One contribution that systematic reviews chronically under-credit is that of the information specialist or research librarian who designs and validates the search. A poorly constructed search undermines a review more surely than almost any other flaw, and a well-constructed one is a genuine methodological achievement. Recording this work explicitly under Methodology and Investigation — rather than relegating it to an acknowledgement — is one of the clearest practical gains from applying contributorship to evidence synthesis. It names a contribution that is both critical and routinely invisible.

    Crediting duplicate work without double-counting

    Reviews rely on tasks done independently by two people — duplicate screening, duplicate extraction — precisely to reduce error. Contributorship should reflect that both reviewers did the work, which CRediT handles naturally by allowing a role to be assigned to more than one contributor. The honest principle, as ever, is that a role records what a person actually did: both screeners earn the Investigation role because both genuinely screened, not as a courtesy. This is the same standard that applies across all contribution recording — credit follows real work, and is neither inflated for visibility nor withheld for convenience.

    A consistent record across systems

    Systematic reviews increasingly register protocols, deposit search strategies and data, and publish in journals that require both PRISMA reporting and a contributorship statement. For that ecosystem to work, the way a contribution is described has to mean the same thing wherever it appears. That consistency is what the CASRAI Dictionary exists to provide: a stable vocabulary so that a Methodology contribution declared in a protocol registry, a manuscript and an institutional record can be recognised as the same claim. Combined with PRISMA’s transparency about process, structured contribution makes the substantial, distributed work of evidence synthesis legible — crediting the screeners, extractors and search designers whose labour holds a review together, not only the names at the top of the list.

  • Authorship in large clinical trials: bylines, groups and the CRediT statement

    A large randomised clinical trial is one of the most collaborative undertakings in research. A single phase III study can involve hundreds of investigators across dozens of sites, plus statisticians, data managers, monitoring committees, and the coordinating centre that holds it all together. Yet the article that reports the result carries a byline that may name only a dozen people. How can a byline that small honestly represent an effort that large? The answer lies in a set of conventions — the ICMJE authorship criteria, group authorship, and a structured contributions statement — that together let a vast, distributed effort be recorded fairly without diluting accountability. This sits squarely in the territory the guidance on group authorship addresses, and it leans heavily on the CRediT roles to do the fine-grained work.

    The ICMJE authorship criteria

    The reference point for who counts as an author of a clinical trial is the International Committee of Medical Journal Editors (ICMJE). The ICMJE recommendations set out four criteria, all of which an individual must meet to qualify for authorship:

    1. Substantial contributions to the conception or design of the work, or the acquisition, analysis, or interpretation of data;
    2. Drafting the work or revising it critically for important intellectual content;
    3. Final approval of the version to be published; and
    4. Agreement to be accountable for all aspects of the work, ensuring that questions about its accuracy or integrity are appropriately investigated and resolved.

    The fourth criterion is the one most often forgotten and the most important: authorship is not only credit, it is responsibility. Crucially, the ICMJE is explicit that contributors who meet criterion 1 but not all four should not be authors — they should be acknowledged instead. This is what prevents a trial byline from simply listing everyone who collected data at a site, and it is also why “gift” authorship of senior figures who did not meet the criteria is a breach of the standard rather than a courtesy.

    Group authorship: naming the collaboration

    The criteria create an obvious tension in large trials. Hundreds of people made genuine, substantial contributions, but they cannot all meet every criterion, and a byline of three hundred names would be unusable. Group authorship is the ICMJE-sanctioned solution. A trial is published under the name of the study group — for example, “the EXAMPLE-2 Trial Investigators” — and the article distinguishes between two categories of people:

    • Named authors who individually meet all four ICMJE criteria and take direct responsibility for the manuscript. These may be listed in the byline alongside the group name, or the group name itself may carry the authorship with the qualifying individuals identified.
    • Collaborators — the wider membership of the study group who contributed substantially to the trial but do not individually meet all four authorship criteria. They are listed by name, typically in an appendix or supplementary file, and are indexed by databases such as PubMed as collaborators rather than authors.

    This is not a way of demoting people. It is a way of telling the truth about a collaboration: a clear, public record that this person was a named author who is accountable for the paper, and that person was a collaborating investigator whose site recruited patients and whose work made the trial possible. The ICMJE asks that the journal specify, somewhere in the article, exactly who is responsible in each category, so that the byline cannot be read as either inflating or erasing anyone’s role.

    The CRediT statement: who did what

    Group authorship answers who is accountable; it does not, by itself, answer who did what. That is the job of a structured contributions statement built on the CRediT taxonomy. CRediT supplies fourteen standard roles — Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Resources, Data curation, Writing – original draft, Writing – review & editing, Visualization, Supervision, Project administration, and Funding acquisition — and a clinical trial maps onto them unusually cleanly:

    • Conceptualization and Methodology capture the people who designed the trial and its protocol.
    • Investigation records the site investigators who recruited and assessed participants — the role that most collaborating investigators occupy.
    • Data curation records the data managers who cleaned and maintained the trial database.
    • Formal analysis records the statisticians who ran the pre-specified analyses.
    • Project administration records the coordinating-centre staff who ran the trial day to day.
    • Funding acquisition records those who secured the grant that paid for it.

    A CRediT statement turns the contributions section from a vague paragraph of thanks into machine-readable data. For a named author it can specify a degree of contribution — lead, equal, or supporting — so that the chief investigator’s role in Conceptualization is distinguished from a co-author’s supporting role. Used well, it means that a reader, a hiring committee, or a CRIS system can see precisely how the labour of a trial was divided, rather than inferring it from author order alone.

    Why this matters for accountability

    Clinical trials carry unusual weight: their results change clinical practice and affect patients. That raises the stakes of the fourth ICMJE criterion. When questions are later raised about a trial — a data discrepancy, an unreproducible analysis, a query about a particular site — the structured record of authorship and contribution is what makes accountability tractable. Knowing who took responsibility as a named author, who curated the data, and who ran the statistics localises responsibility rather than diffusing it across an anonymous group. A trial that is published under a study-group name and carries both a clear statement of named authors and a CRediT breakdown is far more accountable than one with a long, undifferentiated byline.

    Practical guidance for trial teams

    1. Decide authorship against the ICMJE criteria, not seniority or site. Meeting criterion 1 alone makes someone a collaborator, not an author.
    2. Adopt a group name and a written authorship policy early — ideally in the protocol — so that expectations are set before the data exist and disputes are pre-empted.
    3. List collaborators by name. Being a collaborating investigator is a real, indexed, citable contribution; do not bury it.
    4. Publish a CRediT statement mapping each named author to their roles, with degree-of-contribution where it clarifies who led what.
    5. Attach ORCID iDs to named authors and, where possible, collaborators, so that the contribution propagates to each person’s record rather than being lost to a generic group name.

    Where shared vocabulary fits

    “Author”, “collaborator”, “investigator”, “study group”, and “contributor” are used inconsistently across trials and journals, which is part of why trial bylines are so often misread. A shared, federated vocabulary that defines these terms precisely — and points back to the ICMJE recommendations and the CRediT taxonomy — is what lets a group-authored trial in one journal be understood the same way in another. Supplying that definitional layer is the role the CASRAI dictionary is designed to play.

    Related reading